Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = '/data'
!pip install matplotlib==2.0.2
# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Collecting matplotlib==2.0.2
  Downloading https://files.pythonhosted.org/packages/60/d4/6b6d8a7a6bc69a1602ab372f6fc6e88ef88a8a96398a1a25edbac636295b/matplotlib-2.0.2-cp36-cp36m-manylinux1_x86_64.whl (14.6MB)
    100% |████████████████████████████████| 14.6MB 37kB/s  eta 0:00:01
Requirement already satisfied: six>=1.10 in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already satisfied: numpy>=1.7.1 in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already satisfied: python-dateutil in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already satisfied: pytz in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Requirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.6/site-packages/cycler-0.10.0-py3.6.egg (from matplotlib==2.0.2)
Requirement already satisfied: pyparsing!=2.0.0,!=2.0.4,!=2.1.2,!=2.1.6,>=1.5.6 in /opt/conda/lib/python3.6/site-packages (from matplotlib==2.0.2)
Installing collected packages: matplotlib
  Found existing installation: matplotlib 2.1.0
    Uninstalling matplotlib-2.1.0:
      Successfully uninstalled matplotlib-2.1.0
Successfully installed matplotlib-2.0.2
You are using pip version 9.0.1, however version 19.0.3 is available.
You should consider upgrading via the 'pip install --upgrade pip' command.
Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x7fb0c2538588>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7fb0c245b9b0>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.3.0
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    
    real_inputs = tf.placeholder(tf.float32,(None, image_width, image_height, image_channels), "input_real")
    
    z_inputs = tf.placeholder(tf.float32, (None, z_dim), "input_z")
    learning_rate = tf.placeholder(tf.float32, (), "lr")
    
    return real_inputs, z_inputs, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    
    alpha=0.2
    keep_prob=0.9
    
    with tf.variable_scope('discriminator', reuse=reuse):
        
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same', activation=None)
        relu1 = tf.maximum(alpha*x1, x1)
        
        x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same', activation=None)
        bn2 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha*bn2, bn2)
        drop2 = tf.nn.dropout(relu2, keep_prob=keep_prob)

        x3 = tf.layers.conv2d(drop2, 256, 5, strides=2, padding='same')
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha*bn3, bn3)
        drop3 = tf.nn.dropout(relu3, keep_prob=keep_prob)

        flat = tf.reshape(drop3, (-1, 4*4*256))
    
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    
    alpha=0.2
    keep_prob=0.9
    
    with tf.variable_scope('generator', reuse = not is_train):
    
        x1 = tf.layers.dense(z, 7*7*512)
        x1 = tf.reshape(x1, (-1,7,7,512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha*x1, x1)
        x1 = tf.nn.dropout(x1, keep_prob=keep_prob)
        
        x2 = tf.layers.conv2d_transpose(x1, 256, 5, strides=2, padding='same', activation=None)
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha*x2, x2)
        x2 = tf.nn.dropout(x2, keep_prob=keep_prob)
            
        x3 = tf.layers.conv2d_transpose(x2, 128, 5, strides=2, padding='same', activation=None)
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha*x3, x3)
        
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, 3, strides=1, padding='same', activation=None)
        out = tf.tanh(logits)

        return out



"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    smooth = 0.1
    gen = generator(input_z, out_channel_dim, is_train=True)
    d_real, d_logits_real = discriminator(input_real, reuse=False)
    d_fake, d_logits_fake = discriminator(gen, reuse=True)
    
    d_labels_real = tf.ones_like(d_real) * (1 -smooth)
    d_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=d_labels_real))
    
    d_labels_fake = tf.zeros_like(d_fake)
    d_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=d_labels_fake))
    
    d_loss = d_loss_real + d_loss_fake
    g_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_fake)))
    
    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]
    
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train = tf.train.AdamOptimizer(learning_rate=learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train = tf.train.AdamOptimizer(learning_rate=learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)
        
    return d_train, g_train


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    
    input_real, input_z, lr = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    d_loss, g_loss = model_loss(input_real, input_z, data_shape[3])
    d_train, g_train = model_opt(d_loss, g_loss, lr, beta1)

    saver = tf.train.Saver()
    samples,losses = [], []
    step = 0
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                
                # TODO: Train Model
                step += 1
                batch_images *= 2
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                _ = sess.run(d_train, feed_dict={input_real: batch_images, input_z: batch_z, lr: learning_rate})
                _ = sess.run(g_train, feed_dict={input_real: batch_images, input_z: batch_z, lr:learning_rate})
                   
    
                if step % 10 == 0:
                    
                    train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images})
                    train_loss_g = g_loss.eval({input_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                   
                    losses.append((train_loss_d, train_loss_g))


                if step % 100 == 0:
                    show_generator_output(sess, 16, input_z, data_shape[3], data_image_mode)
                    

        saver.save(sess, './generator.ckpt')

                
    return losses

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.3


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 0.4589... Generator Loss: 3.0791
Epoch 1/2... Discriminator Loss: 3.6562... Generator Loss: 0.1055
Epoch 1/2... Discriminator Loss: 0.4903... Generator Loss: 2.2391
Epoch 1/2... Discriminator Loss: 0.3755... Generator Loss: 4.0719
Epoch 1/2... Discriminator Loss: 2.5995... Generator Loss: 7.8311
Epoch 1/2... Discriminator Loss: 1.7490... Generator Loss: 0.7598
Epoch 1/2... Discriminator Loss: 1.6112... Generator Loss: 1.3101
Epoch 1/2... Discriminator Loss: 1.5125... Generator Loss: 1.0278
Epoch 1/2... Discriminator Loss: 1.6311... Generator Loss: 1.2062
Epoch 1/2... Discriminator Loss: 1.5942... Generator Loss: 0.5896

Epoch 1/2... Discriminator Loss: 1.5486... Generator Loss: 0.6620
Epoch 1/2... Discriminator Loss: 1.4170... Generator Loss: 0.6170
Epoch 1/2... Discriminator Loss: 1.5093... Generator Loss: 0.4515
Epoch 1/2... Discriminator Loss: 1.2986... Generator Loss: 0.7279
Epoch 1/2... Discriminator Loss: 1.1772... Generator Loss: 1.0397
Epoch 1/2... Discriminator Loss: 1.6214... Generator Loss: 0.4347
Epoch 1/2... Discriminator Loss: 1.7018... Generator Loss: 0.3536
Epoch 1/2... Discriminator Loss: 1.3305... Generator Loss: 0.5439
Epoch 1/2... Discriminator Loss: 1.1917... Generator Loss: 0.7466
Epoch 1/2... Discriminator Loss: 1.4561... Generator Loss: 1.1585

Epoch 1/2... Discriminator Loss: 1.1762... Generator Loss: 0.7876
Epoch 1/2... Discriminator Loss: 1.1060... Generator Loss: 1.1836
Epoch 1/2... Discriminator Loss: 1.1432... Generator Loss: 1.1996
Epoch 1/2... Discriminator Loss: 1.1565... Generator Loss: 1.2748
Epoch 1/2... Discriminator Loss: 1.3845... Generator Loss: 0.5268
Epoch 1/2... Discriminator Loss: 1.2524... Generator Loss: 0.7204
Epoch 1/2... Discriminator Loss: 1.0824... Generator Loss: 1.8964
Epoch 1/2... Discriminator Loss: 1.4552... Generator Loss: 0.5686
Epoch 1/2... Discriminator Loss: 1.1405... Generator Loss: 1.2759
Epoch 1/2... Discriminator Loss: 0.9027... Generator Loss: 1.1881

Epoch 1/2... Discriminator Loss: 1.3072... Generator Loss: 0.6446
Epoch 1/2... Discriminator Loss: 1.2327... Generator Loss: 0.7260
Epoch 1/2... Discriminator Loss: 0.9609... Generator Loss: 1.5453
Epoch 1/2... Discriminator Loss: 0.9678... Generator Loss: 1.6563
Epoch 1/2... Discriminator Loss: 0.9530... Generator Loss: 1.8966
Epoch 1/2... Discriminator Loss: 1.0779... Generator Loss: 2.1273
Epoch 1/2... Discriminator Loss: 1.2131... Generator Loss: 0.9302
Epoch 1/2... Discriminator Loss: 1.0316... Generator Loss: 1.4170
Epoch 1/2... Discriminator Loss: 1.3670... Generator Loss: 2.3509
Epoch 1/2... Discriminator Loss: 1.5721... Generator Loss: 2.5244

Epoch 1/2... Discriminator Loss: 1.0452... Generator Loss: 1.2294
Epoch 1/2... Discriminator Loss: 1.4036... Generator Loss: 0.6283
Epoch 1/2... Discriminator Loss: 1.5339... Generator Loss: 0.6138
Epoch 1/2... Discriminator Loss: 1.1993... Generator Loss: 1.4475
Epoch 1/2... Discriminator Loss: 1.3580... Generator Loss: 0.5904
Epoch 1/2... Discriminator Loss: 0.9244... Generator Loss: 1.0527
Epoch 1/2... Discriminator Loss: 1.4687... Generator Loss: 1.7435
Epoch 1/2... Discriminator Loss: 1.3057... Generator Loss: 0.7227
Epoch 1/2... Discriminator Loss: 1.2611... Generator Loss: 1.7567
Epoch 1/2... Discriminator Loss: 1.2748... Generator Loss: 0.6932

Epoch 1/2... Discriminator Loss: 1.1953... Generator Loss: 0.6274
Epoch 1/2... Discriminator Loss: 1.3751... Generator Loss: 0.5679
Epoch 1/2... Discriminator Loss: 1.2293... Generator Loss: 0.6650
Epoch 1/2... Discriminator Loss: 1.2680... Generator Loss: 0.8412
Epoch 1/2... Discriminator Loss: 1.1775... Generator Loss: 0.9353
Epoch 1/2... Discriminator Loss: 1.1457... Generator Loss: 0.7855
Epoch 1/2... Discriminator Loss: 1.6109... Generator Loss: 0.3830
Epoch 1/2... Discriminator Loss: 1.3284... Generator Loss: 1.6720
Epoch 1/2... Discriminator Loss: 1.1719... Generator Loss: 0.8852
Epoch 1/2... Discriminator Loss: 1.2203... Generator Loss: 0.6766

Epoch 1/2... Discriminator Loss: 1.2791... Generator Loss: 0.7605
Epoch 1/2... Discriminator Loss: 1.0434... Generator Loss: 0.7497
Epoch 1/2... Discriminator Loss: 1.1268... Generator Loss: 1.6295
Epoch 1/2... Discriminator Loss: 1.2237... Generator Loss: 1.2737
Epoch 1/2... Discriminator Loss: 1.4568... Generator Loss: 0.5377
Epoch 1/2... Discriminator Loss: 1.2260... Generator Loss: 0.6763
Epoch 1/2... Discriminator Loss: 1.1202... Generator Loss: 1.1415
Epoch 1/2... Discriminator Loss: 0.9990... Generator Loss: 1.5881
Epoch 1/2... Discriminator Loss: 1.2546... Generator Loss: 0.6413
Epoch 1/2... Discriminator Loss: 1.4642... Generator Loss: 1.7662

Epoch 1/2... Discriminator Loss: 1.2217... Generator Loss: 0.8560
Epoch 1/2... Discriminator Loss: 1.0619... Generator Loss: 1.3455
Epoch 1/2... Discriminator Loss: 1.2520... Generator Loss: 0.6379
Epoch 1/2... Discriminator Loss: 1.4300... Generator Loss: 0.6351
Epoch 1/2... Discriminator Loss: 1.5102... Generator Loss: 0.4605
Epoch 1/2... Discriminator Loss: 1.4193... Generator Loss: 0.7414
Epoch 1/2... Discriminator Loss: 1.1785... Generator Loss: 0.8451
Epoch 1/2... Discriminator Loss: 1.1123... Generator Loss: 0.6746
Epoch 1/2... Discriminator Loss: 1.2747... Generator Loss: 1.3608
Epoch 1/2... Discriminator Loss: 1.2106... Generator Loss: 0.6473

Epoch 1/2... Discriminator Loss: 1.1523... Generator Loss: 0.8246
Epoch 1/2... Discriminator Loss: 1.1595... Generator Loss: 0.7403
Epoch 1/2... Discriminator Loss: 0.9919... Generator Loss: 1.0213
Epoch 1/2... Discriminator Loss: 1.3117... Generator Loss: 0.7402
Epoch 1/2... Discriminator Loss: 1.1711... Generator Loss: 0.6687
Epoch 1/2... Discriminator Loss: 1.0706... Generator Loss: 0.8217
Epoch 1/2... Discriminator Loss: 1.3643... Generator Loss: 0.4871
Epoch 1/2... Discriminator Loss: 1.2884... Generator Loss: 0.7938
Epoch 1/2... Discriminator Loss: 1.4042... Generator Loss: 0.5652
Epoch 1/2... Discriminator Loss: 1.2322... Generator Loss: 0.7411

Epoch 1/2... Discriminator Loss: 1.3200... Generator Loss: 0.5931
Epoch 1/2... Discriminator Loss: 1.3675... Generator Loss: 1.8543
Epoch 1/2... Discriminator Loss: 1.5790... Generator Loss: 0.4940
Epoch 1/2... Discriminator Loss: 1.2196... Generator Loss: 0.9163
Epoch 1/2... Discriminator Loss: 1.3069... Generator Loss: 0.5428
Epoch 1/2... Discriminator Loss: 1.1942... Generator Loss: 0.9847
Epoch 1/2... Discriminator Loss: 1.0549... Generator Loss: 1.0057
Epoch 1/2... Discriminator Loss: 1.0867... Generator Loss: 0.9209
Epoch 1/2... Discriminator Loss: 1.1987... Generator Loss: 0.5913
Epoch 1/2... Discriminator Loss: 1.3014... Generator Loss: 0.6676

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.3


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 1.3527... Generator Loss: 0.7441
Epoch 1/2... Discriminator Loss: 0.6500... Generator Loss: 1.7216
Epoch 1/2... Discriminator Loss: 0.6013... Generator Loss: 1.9100
Epoch 1/2... Discriminator Loss: 1.6283... Generator Loss: 0.5331
Epoch 1/2... Discriminator Loss: 0.9095... Generator Loss: 1.6827
Epoch 1/2... Discriminator Loss: 1.7799... Generator Loss: 0.3577
Epoch 1/2... Discriminator Loss: 1.8813... Generator Loss: 0.3464
Epoch 1/2... Discriminator Loss: 0.9581... Generator Loss: 0.9220
Epoch 1/2... Discriminator Loss: 0.8006... Generator Loss: 1.3042
Epoch 1/2... Discriminator Loss: 0.7002... Generator Loss: 2.4172

Epoch 1/2... Discriminator Loss: 1.0582... Generator Loss: 0.7849
Epoch 1/2... Discriminator Loss: 0.7550... Generator Loss: 1.4113
Epoch 1/2... Discriminator Loss: 0.8468... Generator Loss: 1.0669
Epoch 1/2... Discriminator Loss: 1.0375... Generator Loss: 0.9341
Epoch 1/2... Discriminator Loss: 1.8932... Generator Loss: 0.3409
Epoch 1/2... Discriminator Loss: 1.1923... Generator Loss: 2.6641
Epoch 1/2... Discriminator Loss: 0.6801... Generator Loss: 2.3187
Epoch 1/2... Discriminator Loss: 0.6967... Generator Loss: 1.8098
Epoch 1/2... Discriminator Loss: 1.5364... Generator Loss: 0.5643
Epoch 1/2... Discriminator Loss: 1.1817... Generator Loss: 0.6174

Epoch 1/2... Discriminator Loss: 0.7422... Generator Loss: 1.5077
Epoch 1/2... Discriminator Loss: 1.0543... Generator Loss: 2.8390
Epoch 1/2... Discriminator Loss: 1.1233... Generator Loss: 0.8370
Epoch 1/2... Discriminator Loss: 1.4078... Generator Loss: 0.6942
Epoch 1/2... Discriminator Loss: 1.9338... Generator Loss: 0.2998
Epoch 1/2... Discriminator Loss: 1.7293... Generator Loss: 0.3695
Epoch 1/2... Discriminator Loss: 0.8419... Generator Loss: 1.3718
Epoch 1/2... Discriminator Loss: 0.9122... Generator Loss: 1.0264
Epoch 1/2... Discriminator Loss: 0.5779... Generator Loss: 3.2783
Epoch 1/2... Discriminator Loss: 0.6318... Generator Loss: 2.4581

Epoch 1/2... Discriminator Loss: 2.1108... Generator Loss: 0.2441
Epoch 1/2... Discriminator Loss: 0.8792... Generator Loss: 1.2130
Epoch 1/2... Discriminator Loss: 1.1945... Generator Loss: 0.6938
Epoch 1/2... Discriminator Loss: 0.6835... Generator Loss: 1.5153
Epoch 1/2... Discriminator Loss: 1.4790... Generator Loss: 0.5202
Epoch 1/2... Discriminator Loss: 0.9241... Generator Loss: 3.5050
Epoch 1/2... Discriminator Loss: 0.8672... Generator Loss: 1.3024
Epoch 1/2... Discriminator Loss: 1.3641... Generator Loss: 0.6909
Epoch 1/2... Discriminator Loss: 1.2235... Generator Loss: 0.6543
Epoch 1/2... Discriminator Loss: 0.7534... Generator Loss: 3.3822

Epoch 1/2... Discriminator Loss: 0.5681... Generator Loss: 2.4238
Epoch 1/2... Discriminator Loss: 1.4127... Generator Loss: 0.4661
Epoch 1/2... Discriminator Loss: 0.6072... Generator Loss: 3.3215
Epoch 1/2... Discriminator Loss: 1.1318... Generator Loss: 0.8115
Epoch 1/2... Discriminator Loss: 1.4435... Generator Loss: 2.5940
Epoch 1/2... Discriminator Loss: 0.8799... Generator Loss: 1.5371
Epoch 1/2... Discriminator Loss: 1.2078... Generator Loss: 1.8321
Epoch 1/2... Discriminator Loss: 1.1727... Generator Loss: 0.7344
Epoch 1/2... Discriminator Loss: 1.1310... Generator Loss: 0.9048
Epoch 1/2... Discriminator Loss: 0.6761... Generator Loss: 1.6107

Epoch 1/2... Discriminator Loss: 1.8707... Generator Loss: 0.3352
Epoch 1/2... Discriminator Loss: 2.2066... Generator Loss: 0.1834
Epoch 1/2... Discriminator Loss: 1.5358... Generator Loss: 1.2983
Epoch 1/2... Discriminator Loss: 1.3072... Generator Loss: 0.5868
Epoch 1/2... Discriminator Loss: 0.5403... Generator Loss: 2.0968
Epoch 1/2... Discriminator Loss: 0.9884... Generator Loss: 3.1169
Epoch 1/2... Discriminator Loss: 0.7642... Generator Loss: 1.5506
Epoch 1/2... Discriminator Loss: 2.0217... Generator Loss: 1.4728
Epoch 1/2... Discriminator Loss: 1.0314... Generator Loss: 0.8396
Epoch 1/2... Discriminator Loss: 0.6029... Generator Loss: 1.8941

Epoch 1/2... Discriminator Loss: 1.3876... Generator Loss: 1.0618
Epoch 1/2... Discriminator Loss: 0.6830... Generator Loss: 1.9699
Epoch 1/2... Discriminator Loss: 1.0601... Generator Loss: 0.8208
Epoch 1/2... Discriminator Loss: 0.6464... Generator Loss: 1.8103
Epoch 1/2... Discriminator Loss: 0.8556... Generator Loss: 1.1541
Epoch 1/2... Discriminator Loss: 0.5606... Generator Loss: 3.2626
Epoch 1/2... Discriminator Loss: 0.8550... Generator Loss: 1.0614
Epoch 1/2... Discriminator Loss: 0.5503... Generator Loss: 2.1624
Epoch 1/2... Discriminator Loss: 1.6552... Generator Loss: 0.4897
Epoch 1/2... Discriminator Loss: 1.2896... Generator Loss: 1.1475

Epoch 1/2... Discriminator Loss: 1.2042... Generator Loss: 0.8219
Epoch 1/2... Discriminator Loss: 0.8898... Generator Loss: 1.2187
Epoch 1/2... Discriminator Loss: 1.1474... Generator Loss: 0.7876
Epoch 1/2... Discriminator Loss: 0.8711... Generator Loss: 0.9760
Epoch 1/2... Discriminator Loss: 1.0372... Generator Loss: 0.7997
Epoch 1/2... Discriminator Loss: 1.3380... Generator Loss: 1.0409
Epoch 1/2... Discriminator Loss: 0.4542... Generator Loss: 3.0485
Epoch 1/2... Discriminator Loss: 0.8223... Generator Loss: 1.1643
Epoch 1/2... Discriminator Loss: 3.9142... Generator Loss: 4.1490
Epoch 1/2... Discriminator Loss: 0.6943... Generator Loss: 2.8249

Epoch 1/2... Discriminator Loss: 0.8676... Generator Loss: 1.1661
Epoch 1/2... Discriminator Loss: 1.0039... Generator Loss: 0.9454
Epoch 1/2... Discriminator Loss: 1.8738... Generator Loss: 1.9135
Epoch 1/2... Discriminator Loss: 0.7178... Generator Loss: 2.8565
Epoch 1/2... Discriminator Loss: 1.1223... Generator Loss: 0.8676
Epoch 1/2... Discriminator Loss: 0.9367... Generator Loss: 1.1658
Epoch 1/2... Discriminator Loss: 1.1149... Generator Loss: 0.8353
Epoch 1/2... Discriminator Loss: 0.7906... Generator Loss: 1.1470
Epoch 1/2... Discriminator Loss: 1.7885... Generator Loss: 0.7965
Epoch 1/2... Discriminator Loss: 0.5699... Generator Loss: 2.3957

Epoch 1/2... Discriminator Loss: 0.9122... Generator Loss: 1.4699
Epoch 1/2... Discriminator Loss: 1.2174... Generator Loss: 0.6834
Epoch 1/2... Discriminator Loss: 0.9083... Generator Loss: 1.8147
Epoch 1/2... Discriminator Loss: 0.4475... Generator Loss: 3.3473
Epoch 1/2... Discriminator Loss: 0.6174... Generator Loss: 1.7621
Epoch 1/2... Discriminator Loss: 0.5506... Generator Loss: 1.7518
Epoch 1/2... Discriminator Loss: 0.5309... Generator Loss: 2.7193
Epoch 1/2... Discriminator Loss: 1.3450... Generator Loss: 0.9057
Epoch 1/2... Discriminator Loss: 0.8162... Generator Loss: 1.0980
Epoch 1/2... Discriminator Loss: 0.6956... Generator Loss: 1.5223

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 2.0166... Generator Loss: 0.4364
Epoch 1/2... Discriminator Loss: 0.6982... Generator Loss: 2.5163
Epoch 1/2... Discriminator Loss: 1.0258... Generator Loss: 3.3628
Epoch 1/2... Discriminator Loss: 1.4318... Generator Loss: 0.6644
Epoch 1/2... Discriminator Loss: 1.0078... Generator Loss: 1.0512
Epoch 1/2... Discriminator Loss: 2.1052... Generator Loss: 0.2975
Epoch 1/2... Discriminator Loss: 2.4858... Generator Loss: 0.2171
Epoch 1/2... Discriminator Loss: 0.9404... Generator Loss: 0.9805
Epoch 1/2... Discriminator Loss: 2.0375... Generator Loss: 0.2768
Epoch 1/2... Discriminator Loss: 2.5688... Generator Loss: 0.1460

Epoch 1/2... Discriminator Loss: 1.3907... Generator Loss: 0.6475
Epoch 1/2... Discriminator Loss: 1.9533... Generator Loss: 0.3060
Epoch 1/2... Discriminator Loss: 3.3058... Generator Loss: 0.0757
Epoch 1/2... Discriminator Loss: 1.3437... Generator Loss: 0.6116
Epoch 1/2... Discriminator Loss: 1.2575... Generator Loss: 0.6956
Epoch 1/2... Discriminator Loss: 1.2362... Generator Loss: 0.6424
Epoch 1/2... Discriminator Loss: 1.4457... Generator Loss: 0.5040
Epoch 1/2... Discriminator Loss: 2.0462... Generator Loss: 0.2125
Epoch 1/2... Discriminator Loss: 0.7130... Generator Loss: 1.3616
Epoch 1/2... Discriminator Loss: 1.0360... Generator Loss: 1.8435

Epoch 1/2... Discriminator Loss: 1.1506... Generator Loss: 2.3218
Epoch 1/2... Discriminator Loss: 0.7845... Generator Loss: 2.5990
Epoch 1/2... Discriminator Loss: 0.8335... Generator Loss: 1.2121
Epoch 1/2... Discriminator Loss: 0.7308... Generator Loss: 2.5758
Epoch 1/2... Discriminator Loss: 1.0561... Generator Loss: 0.8768
Epoch 1/2... Discriminator Loss: 1.7047... Generator Loss: 0.3823
Epoch 1/2... Discriminator Loss: 1.1404... Generator Loss: 0.7683
Epoch 1/2... Discriminator Loss: 0.8467... Generator Loss: 1.2830
Epoch 1/2... Discriminator Loss: 0.7466... Generator Loss: 1.8383
Epoch 1/2... Discriminator Loss: 1.9800... Generator Loss: 2.5337

Epoch 1/2... Discriminator Loss: 1.5230... Generator Loss: 0.4040
Epoch 1/2... Discriminator Loss: 1.3611... Generator Loss: 0.5703
Epoch 1/2... Discriminator Loss: 0.9876... Generator Loss: 1.0249
Epoch 1/2... Discriminator Loss: 0.6078... Generator Loss: 1.9969
Epoch 1/2... Discriminator Loss: 0.8937... Generator Loss: 1.0951
Epoch 1/2... Discriminator Loss: 0.9791... Generator Loss: 0.8712
Epoch 1/2... Discriminator Loss: 1.2562... Generator Loss: 4.2067
Epoch 1/2... Discriminator Loss: 0.6960... Generator Loss: 1.8532
Epoch 1/2... Discriminator Loss: 1.6181... Generator Loss: 0.4195
Epoch 1/2... Discriminator Loss: 1.5603... Generator Loss: 2.4279

Epoch 1/2... Discriminator Loss: 1.5338... Generator Loss: 0.5969
Epoch 1/2... Discriminator Loss: 1.3944... Generator Loss: 0.5575
Epoch 1/2... Discriminator Loss: 0.8768... Generator Loss: 1.1876
Epoch 1/2... Discriminator Loss: 1.0844... Generator Loss: 2.5677
Epoch 1/2... Discriminator Loss: 0.9456... Generator Loss: 0.9317
Epoch 1/2... Discriminator Loss: 1.4883... Generator Loss: 0.5020
Epoch 1/2... Discriminator Loss: 1.4937... Generator Loss: 0.8233
Epoch 1/2... Discriminator Loss: 0.6907... Generator Loss: 2.8583
Epoch 1/2... Discriminator Loss: 1.1233... Generator Loss: 0.8716
Epoch 1/2... Discriminator Loss: 0.6715... Generator Loss: 3.7560

Epoch 1/2... Discriminator Loss: 0.9191... Generator Loss: 2.5438
Epoch 1/2... Discriminator Loss: 0.5599... Generator Loss: 1.9895
Epoch 1/2... Discriminator Loss: 0.8987... Generator Loss: 2.8158
Epoch 1/2... Discriminator Loss: 1.2453... Generator Loss: 3.0998
Epoch 1/2... Discriminator Loss: 1.7771... Generator Loss: 0.4040
Epoch 1/2... Discriminator Loss: 0.9524... Generator Loss: 1.4999
Epoch 1/2... Discriminator Loss: 0.9101... Generator Loss: 1.2865
Epoch 1/2... Discriminator Loss: 0.6155... Generator Loss: 1.8692
Epoch 1/2... Discriminator Loss: 1.5040... Generator Loss: 0.4508
Epoch 1/2... Discriminator Loss: 2.2970... Generator Loss: 2.1093

Epoch 1/2... Discriminator Loss: 0.8533... Generator Loss: 1.2951
Epoch 1/2... Discriminator Loss: 1.1270... Generator Loss: 0.7502
Epoch 1/2... Discriminator Loss: 0.6267... Generator Loss: 1.9266
Epoch 1/2... Discriminator Loss: 2.0158... Generator Loss: 2.6111
Epoch 1/2... Discriminator Loss: 1.2085... Generator Loss: 0.7033
Epoch 1/2... Discriminator Loss: 1.0323... Generator Loss: 0.8442
Epoch 1/2... Discriminator Loss: 0.6658... Generator Loss: 2.0810
Epoch 1/2... Discriminator Loss: 0.7957... Generator Loss: 1.3439
Epoch 1/2... Discriminator Loss: 1.3231... Generator Loss: 1.0977
Epoch 1/2... Discriminator Loss: 1.4054... Generator Loss: 1.8909

Epoch 1/2... Discriminator Loss: 1.2333... Generator Loss: 0.6270
Epoch 1/2... Discriminator Loss: 1.2128... Generator Loss: 1.2418
Epoch 1/2... Discriminator Loss: 1.1702... Generator Loss: 0.8703
Epoch 1/2... Discriminator Loss: 1.3768... Generator Loss: 0.5722
Epoch 1/2... Discriminator Loss: 2.2077... Generator Loss: 3.0143
Epoch 1/2... Discriminator Loss: 0.9767... Generator Loss: 2.0079
Epoch 1/2... Discriminator Loss: 1.4748... Generator Loss: 2.0647
Epoch 1/2... Discriminator Loss: 1.4977... Generator Loss: 2.0809
Epoch 1/2... Discriminator Loss: 0.4953... Generator Loss: 2.6955
Epoch 1/2... Discriminator Loss: 1.2281... Generator Loss: 1.2242

Epoch 1/2... Discriminator Loss: 2.2422... Generator Loss: 1.9472
Epoch 1/2... Discriminator Loss: 1.1504... Generator Loss: 1.2224
Epoch 1/2... Discriminator Loss: 0.5901... Generator Loss: 1.7237
Epoch 1/2... Discriminator Loss: 1.4138... Generator Loss: 1.3085
Epoch 1/2... Discriminator Loss: 0.7949... Generator Loss: 1.7070
Epoch 1/2... Discriminator Loss: 0.9467... Generator Loss: 2.4434
Epoch 1/2... Discriminator Loss: 1.3614... Generator Loss: 1.1842
Epoch 1/2... Discriminator Loss: 1.3289... Generator Loss: 0.5495
Epoch 1/2... Discriminator Loss: 1.0355... Generator Loss: 0.9564
Epoch 1/2... Discriminator Loss: 1.6397... Generator Loss: 0.3470

Epoch 1/2... Discriminator Loss: 0.7930... Generator Loss: 2.3506
Epoch 1/2... Discriminator Loss: 1.1628... Generator Loss: 0.7457
Epoch 1/2... Discriminator Loss: 1.5136... Generator Loss: 0.4051
Epoch 1/2... Discriminator Loss: 1.8665... Generator Loss: 2.2833
Epoch 1/2... Discriminator Loss: 1.1116... Generator Loss: 0.6957
Epoch 1/2... Discriminator Loss: 0.5456... Generator Loss: 2.6918
Epoch 1/2... Discriminator Loss: 1.3973... Generator Loss: 0.6848
Epoch 1/2... Discriminator Loss: 0.5441... Generator Loss: 2.1849
Epoch 1/2... Discriminator Loss: 1.1073... Generator Loss: 2.8256
Epoch 1/2... Discriminator Loss: 1.2364... Generator Loss: 3.2382

batch_size = 32
z_dim = 100
learning_rate = 0.001
beta1 = 0.2


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 1.7179... Generator Loss: 26.4269
Epoch 1/2... Discriminator Loss: 0.9482... Generator Loss: 4.0883
Epoch 1/2... Discriminator Loss: 1.0483... Generator Loss: 1.3689
Epoch 1/2... Discriminator Loss: 2.6138... Generator Loss: 0.1459
Epoch 1/2... Discriminator Loss: 3.1652... Generator Loss: 0.0748
Epoch 1/2... Discriminator Loss: 1.7501... Generator Loss: 0.4773
Epoch 1/2... Discriminator Loss: 3.3167... Generator Loss: 0.0693
Epoch 1/2... Discriminator Loss: 0.5805... Generator Loss: 2.9984
Epoch 1/2... Discriminator Loss: 0.9405... Generator Loss: 2.3544
Epoch 1/2... Discriminator Loss: 1.5017... Generator Loss: 3.0239

Epoch 1/2... Discriminator Loss: 2.4269... Generator Loss: 0.2024
Epoch 1/2... Discriminator Loss: 1.0705... Generator Loss: 0.9655
Epoch 1/2... Discriminator Loss: 2.1408... Generator Loss: 0.2527
Epoch 1/2... Discriminator Loss: 0.8613... Generator Loss: 1.5097
Epoch 1/2... Discriminator Loss: 1.2212... Generator Loss: 1.1845
Epoch 1/2... Discriminator Loss: 1.4923... Generator Loss: 2.1164
Epoch 1/2... Discriminator Loss: 1.3272... Generator Loss: 0.7777
Epoch 1/2... Discriminator Loss: 1.1602... Generator Loss: 0.8632
Epoch 1/2... Discriminator Loss: 1.5221... Generator Loss: 0.4658
Epoch 1/2... Discriminator Loss: 1.5327... Generator Loss: 0.9846

Epoch 1/2... Discriminator Loss: 1.0118... Generator Loss: 0.9811
Epoch 1/2... Discriminator Loss: 1.6099... Generator Loss: 0.7824
Epoch 1/2... Discriminator Loss: 2.6191... Generator Loss: 0.1110
Epoch 1/2... Discriminator Loss: 1.4571... Generator Loss: 0.7620
Epoch 1/2... Discriminator Loss: 1.4411... Generator Loss: 0.7429
Epoch 1/2... Discriminator Loss: 1.2356... Generator Loss: 1.5294
Epoch 1/2... Discriminator Loss: 1.6053... Generator Loss: 0.4331
Epoch 1/2... Discriminator Loss: 1.3044... Generator Loss: 0.9896
Epoch 1/2... Discriminator Loss: 2.5426... Generator Loss: 0.1547
Epoch 1/2... Discriminator Loss: 1.5777... Generator Loss: 0.8271

Epoch 1/2... Discriminator Loss: 1.3850... Generator Loss: 0.8445
Epoch 1/2... Discriminator Loss: 1.2311... Generator Loss: 0.8114
Epoch 1/2... Discriminator Loss: 1.4686... Generator Loss: 1.0981
Epoch 1/2... Discriminator Loss: 1.3555... Generator Loss: 0.9314
Epoch 1/2... Discriminator Loss: 1.5281... Generator Loss: 0.5026
Epoch 1/2... Discriminator Loss: 1.0037... Generator Loss: 0.8586
Epoch 1/2... Discriminator Loss: 1.6165... Generator Loss: 0.6323
Epoch 1/2... Discriminator Loss: 1.4235... Generator Loss: 0.6514
Epoch 1/2... Discriminator Loss: 1.2969... Generator Loss: 0.7998
Epoch 1/2... Discriminator Loss: 1.3335... Generator Loss: 0.9500

Epoch 1/2... Discriminator Loss: 1.2750... Generator Loss: 0.7721
Epoch 1/2... Discriminator Loss: 1.4433... Generator Loss: 0.5790
Epoch 1/2... Discriminator Loss: 1.5728... Generator Loss: 0.4872
Epoch 1/2... Discriminator Loss: 1.6497... Generator Loss: 0.4639
Epoch 1/2... Discriminator Loss: 1.1213... Generator Loss: 0.9014
Epoch 1/2... Discriminator Loss: 1.4079... Generator Loss: 0.6961
Epoch 1/2... Discriminator Loss: 1.2629... Generator Loss: 0.8848
Epoch 1/2... Discriminator Loss: 1.2094... Generator Loss: 1.1568
Epoch 1/2... Discriminator Loss: 1.3540... Generator Loss: 0.8882
Epoch 1/2... Discriminator Loss: 1.4936... Generator Loss: 1.0978

Epoch 1/2... Discriminator Loss: 1.4912... Generator Loss: 0.6236
Epoch 1/2... Discriminator Loss: 1.4861... Generator Loss: 0.5427
Epoch 1/2... Discriminator Loss: 1.3876... Generator Loss: 1.1897
Epoch 1/2... Discriminator Loss: 1.5223... Generator Loss: 0.6986
Epoch 1/2... Discriminator Loss: 1.1881... Generator Loss: 1.0448
Epoch 1/2... Discriminator Loss: 2.1608... Generator Loss: 2.2534
Epoch 1/2... Discriminator Loss: 1.4331... Generator Loss: 0.9134
Epoch 1/2... Discriminator Loss: 1.7085... Generator Loss: 1.4877
Epoch 1/2... Discriminator Loss: 1.2545... Generator Loss: 0.8572
Epoch 1/2... Discriminator Loss: 1.3980... Generator Loss: 0.7508

Epoch 1/2... Discriminator Loss: 1.3159... Generator Loss: 0.5197
Epoch 1/2... Discriminator Loss: 1.3346... Generator Loss: 0.6377
Epoch 1/2... Discriminator Loss: 1.3013... Generator Loss: 0.6229
Epoch 1/2... Discriminator Loss: 1.4351... Generator Loss: 0.8761
Epoch 1/2... Discriminator Loss: 1.5480... Generator Loss: 0.4092
Epoch 1/2... Discriminator Loss: 1.1904... Generator Loss: 0.7526
Epoch 1/2... Discriminator Loss: 1.5130... Generator Loss: 0.6242
Epoch 1/2... Discriminator Loss: 1.2769... Generator Loss: 0.9752
Epoch 1/2... Discriminator Loss: 1.3304... Generator Loss: 0.9878
Epoch 1/2... Discriminator Loss: 1.4335... Generator Loss: 0.5215

Epoch 1/2... Discriminator Loss: 1.2633... Generator Loss: 0.7538
Epoch 1/2... Discriminator Loss: 1.1123... Generator Loss: 0.8638
Epoch 1/2... Discriminator Loss: 1.1216... Generator Loss: 1.2804
Epoch 1/2... Discriminator Loss: 1.4464... Generator Loss: 0.9294
Epoch 1/2... Discriminator Loss: 1.4414... Generator Loss: 1.3500
Epoch 1/2... Discriminator Loss: 1.3563... Generator Loss: 0.6279
Epoch 1/2... Discriminator Loss: 1.3439... Generator Loss: 0.8668
Epoch 1/2... Discriminator Loss: 1.0186... Generator Loss: 1.1933
Epoch 1/2... Discriminator Loss: 1.3569... Generator Loss: 0.8958
Epoch 1/2... Discriminator Loss: 1.4364... Generator Loss: 0.7126

Epoch 1/2... Discriminator Loss: 1.2944... Generator Loss: 0.9423
Epoch 1/2... Discriminator Loss: 1.3559... Generator Loss: 0.6481
Epoch 1/2... Discriminator Loss: 1.5721... Generator Loss: 0.5008
Epoch 1/2... Discriminator Loss: 0.8045... Generator Loss: 1.0043
Epoch 1/2... Discriminator Loss: 1.2762... Generator Loss: 1.3375
Epoch 1/2... Discriminator Loss: 1.4631... Generator Loss: 0.7320
Epoch 1/2... Discriminator Loss: 1.2009... Generator Loss: 0.9841
Epoch 1/2... Discriminator Loss: 1.3610... Generator Loss: 0.6881
Epoch 1/2... Discriminator Loss: 1.3655... Generator Loss: 0.8642
Epoch 1/2... Discriminator Loss: 1.4923... Generator Loss: 0.8685

Epoch 1/2... Discriminator Loss: 1.2658... Generator Loss: 0.8729
Epoch 1/2... Discriminator Loss: 1.2643... Generator Loss: 0.6846
Epoch 1/2... Discriminator Loss: 1.0116... Generator Loss: 0.8599
Epoch 1/2... Discriminator Loss: 1.3315... Generator Loss: 0.8003
Epoch 1/2... Discriminator Loss: 1.2836... Generator Loss: 0.5822
Epoch 1/2... Discriminator Loss: 1.4052... Generator Loss: 0.9255
Epoch 1/2... Discriminator Loss: 1.3386... Generator Loss: 0.8951
Epoch 1/2... Discriminator Loss: 1.2997... Generator Loss: 1.4690
Epoch 1/2... Discriminator Loss: 1.4955... Generator Loss: 0.7367
Epoch 1/2... Discriminator Loss: 1.3541... Generator Loss: 1.0692

batch_size = 16
z_dim = 100
learning_rate = 0.001
beta1 = 0.2


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 0.5675... Generator Loss: 15.5801
Epoch 1/2... Discriminator Loss: 2.7110... Generator Loss: 5.0374
Epoch 1/2... Discriminator Loss: 4.8705... Generator Loss: 0.0644
Epoch 1/2... Discriminator Loss: 1.4928... Generator Loss: 2.3555
Epoch 1/2... Discriminator Loss: 2.3351... Generator Loss: 2.7077
Epoch 1/2... Discriminator Loss: 1.8462... Generator Loss: 2.7871
Epoch 1/2... Discriminator Loss: 0.9796... Generator Loss: 1.7096
Epoch 1/2... Discriminator Loss: 1.0846... Generator Loss: 2.5849
Epoch 1/2... Discriminator Loss: 2.9050... Generator Loss: 0.1248
Epoch 1/2... Discriminator Loss: 0.8685... Generator Loss: 1.0832

Epoch 1/2... Discriminator Loss: 0.5902... Generator Loss: 1.6195
Epoch 1/2... Discriminator Loss: 0.6895... Generator Loss: 2.1254
Epoch 1/2... Discriminator Loss: 0.7149... Generator Loss: 1.5473
Epoch 1/2... Discriminator Loss: 1.6361... Generator Loss: 2.1591
Epoch 1/2... Discriminator Loss: 1.6564... Generator Loss: 0.3892
Epoch 1/2... Discriminator Loss: 1.3357... Generator Loss: 0.6730
Epoch 1/2... Discriminator Loss: 1.0974... Generator Loss: 1.8277
Epoch 1/2... Discriminator Loss: 1.8530... Generator Loss: 0.5033
Epoch 1/2... Discriminator Loss: 1.2391... Generator Loss: 0.8451
Epoch 1/2... Discriminator Loss: 1.6802... Generator Loss: 0.5640

Epoch 1/2... Discriminator Loss: 0.6223... Generator Loss: 1.5565
Epoch 1/2... Discriminator Loss: 2.3627... Generator Loss: 0.3045
Epoch 1/2... Discriminator Loss: 1.3455... Generator Loss: 0.8911
Epoch 1/2... Discriminator Loss: 1.5775... Generator Loss: 0.6267
Epoch 1/2... Discriminator Loss: 1.3449... Generator Loss: 1.8045
Epoch 1/2... Discriminator Loss: 0.9559... Generator Loss: 0.9568
Epoch 1/2... Discriminator Loss: 1.3221... Generator Loss: 0.8378
Epoch 1/2... Discriminator Loss: 1.5676... Generator Loss: 0.6504
Epoch 1/2... Discriminator Loss: 1.4271... Generator Loss: 1.2653
Epoch 1/2... Discriminator Loss: 1.5670... Generator Loss: 0.9777

Epoch 1/2... Discriminator Loss: 1.7322... Generator Loss: 0.5342
Epoch 1/2... Discriminator Loss: 2.0680... Generator Loss: 0.3621
Epoch 1/2... Discriminator Loss: 1.5544... Generator Loss: 0.9483
Epoch 1/2... Discriminator Loss: 1.3074... Generator Loss: 0.7240
Epoch 1/2... Discriminator Loss: 1.2485... Generator Loss: 0.7411
Epoch 1/2... Discriminator Loss: 1.2285... Generator Loss: 0.9160
Epoch 1/2... Discriminator Loss: 1.3881... Generator Loss: 0.5768
Epoch 1/2... Discriminator Loss: 1.6308... Generator Loss: 0.9946
Epoch 1/2... Discriminator Loss: 1.1517... Generator Loss: 1.0792
Epoch 1/2... Discriminator Loss: 1.2953... Generator Loss: 0.5210

Epoch 1/2... Discriminator Loss: 1.5931... Generator Loss: 0.7604
Epoch 1/2... Discriminator Loss: 1.4537... Generator Loss: 0.6698
Epoch 1/2... Discriminator Loss: 1.5201... Generator Loss: 0.6936
Epoch 1/2... Discriminator Loss: 1.5688... Generator Loss: 0.7550
Epoch 1/2... Discriminator Loss: 1.1569... Generator Loss: 0.9013
Epoch 1/2... Discriminator Loss: 1.4506... Generator Loss: 0.5527
Epoch 1/2... Discriminator Loss: 1.4348... Generator Loss: 0.9927
Epoch 1/2... Discriminator Loss: 1.4547... Generator Loss: 0.5993
Epoch 1/2... Discriminator Loss: 1.3788... Generator Loss: 1.0860
Epoch 1/2... Discriminator Loss: 1.5286... Generator Loss: 0.5522

Epoch 1/2... Discriminator Loss: 1.1793... Generator Loss: 0.6889
Epoch 1/2... Discriminator Loss: 1.0319... Generator Loss: 1.0701
Epoch 1/2... Discriminator Loss: 1.3454... Generator Loss: 0.9401
Epoch 1/2... Discriminator Loss: 1.2949... Generator Loss: 1.6419
Epoch 1/2... Discriminator Loss: 1.4743... Generator Loss: 0.8071
Epoch 1/2... Discriminator Loss: 1.4370... Generator Loss: 1.3788
Epoch 1/2... Discriminator Loss: 1.3610... Generator Loss: 0.7119
Epoch 1/2... Discriminator Loss: 1.6114... Generator Loss: 1.5782
Epoch 1/2... Discriminator Loss: 1.5801... Generator Loss: 0.7973
Epoch 1/2... Discriminator Loss: 1.3670... Generator Loss: 0.5259

Epoch 1/2... Discriminator Loss: 1.2818... Generator Loss: 0.5838
Epoch 1/2... Discriminator Loss: 1.7269... Generator Loss: 0.7662
Epoch 1/2... Discriminator Loss: 1.3262... Generator Loss: 0.6610
Epoch 1/2... Discriminator Loss: 1.4133... Generator Loss: 0.6461
Epoch 1/2... Discriminator Loss: 1.4284... Generator Loss: 0.5123
Epoch 1/2... Discriminator Loss: 1.4586... Generator Loss: 0.7028
Epoch 1/2... Discriminator Loss: 1.3939... Generator Loss: 1.3144
Epoch 1/2... Discriminator Loss: 1.2436... Generator Loss: 0.9785
Epoch 1/2... Discriminator Loss: 1.2642... Generator Loss: 0.9559
Epoch 1/2... Discriminator Loss: 1.1164... Generator Loss: 1.2600

Epoch 1/2... Discriminator Loss: 1.3774... Generator Loss: 0.9616
Epoch 1/2... Discriminator Loss: 1.3989... Generator Loss: 0.7037
Epoch 1/2... Discriminator Loss: 1.1737... Generator Loss: 0.6201
Epoch 1/2... Discriminator Loss: 1.3224... Generator Loss: 1.1705
Epoch 1/2... Discriminator Loss: 1.4182... Generator Loss: 1.4305
Epoch 1/2... Discriminator Loss: 1.4879... Generator Loss: 0.6920
Epoch 1/2... Discriminator Loss: 1.3787... Generator Loss: 0.4832
Epoch 1/2... Discriminator Loss: 1.3522... Generator Loss: 0.6703
Epoch 1/2... Discriminator Loss: 1.3690... Generator Loss: 0.9433
Epoch 1/2... Discriminator Loss: 1.3746... Generator Loss: 0.7241

Epoch 1/2... Discriminator Loss: 1.3792... Generator Loss: 0.8301
Epoch 1/2... Discriminator Loss: 1.4481... Generator Loss: 0.7604
Epoch 1/2... Discriminator Loss: 1.4482... Generator Loss: 0.7274
Epoch 1/2... Discriminator Loss: 1.4084... Generator Loss: 0.5873
Epoch 1/2... Discriminator Loss: 1.1728... Generator Loss: 0.8538
Epoch 1/2... Discriminator Loss: 0.9287... Generator Loss: 1.7442
Epoch 1/2... Discriminator Loss: 1.1125... Generator Loss: 0.8542
Epoch 1/2... Discriminator Loss: 1.5204... Generator Loss: 0.8937
Epoch 1/2... Discriminator Loss: 1.2981... Generator Loss: 0.8008
Epoch 1/2... Discriminator Loss: 1.5301... Generator Loss: 0.6083

Epoch 1/2... Discriminator Loss: 1.4003... Generator Loss: 0.6683
Epoch 1/2... Discriminator Loss: 1.3284... Generator Loss: 0.8434
Epoch 1/2... Discriminator Loss: 1.5851... Generator Loss: 0.3949
Epoch 1/2... Discriminator Loss: 1.2549... Generator Loss: 1.1785
Epoch 1/2... Discriminator Loss: 1.1722... Generator Loss: 0.9555
Epoch 1/2... Discriminator Loss: 1.3532... Generator Loss: 0.9718
Epoch 1/2... Discriminator Loss: 1.4313... Generator Loss: 0.6084
Epoch 1/2... Discriminator Loss: 1.4426... Generator Loss: 0.6558
Epoch 1/2... Discriminator Loss: 1.1309... Generator Loss: 0.7201
Epoch 1/2... Discriminator Loss: 1.3340... Generator Loss: 0.9320

batch_size = 64
z_dim = 100
learning_rate = 0.001
beta1 = 0.2


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 0.5439... Generator Loss: 25.7220
Epoch 1/2... Discriminator Loss: 0.7656... Generator Loss: 1.9670
Epoch 1/2... Discriminator Loss: 1.8957... Generator Loss: 4.6058
Epoch 1/2... Discriminator Loss: 1.1702... Generator Loss: 1.8811
Epoch 1/2... Discriminator Loss: 0.9678... Generator Loss: 1.6144
Epoch 1/2... Discriminator Loss: 1.0525... Generator Loss: 1.1923
Epoch 1/2... Discriminator Loss: 1.2298... Generator Loss: 4.0794
Epoch 1/2... Discriminator Loss: 2.6647... Generator Loss: 0.1314
Epoch 1/2... Discriminator Loss: 1.2150... Generator Loss: 0.6334
Epoch 1/2... Discriminator Loss: 0.9618... Generator Loss: 1.7295

Epoch 1/2... Discriminator Loss: 0.8432... Generator Loss: 1.7344
Epoch 1/2... Discriminator Loss: 0.8117... Generator Loss: 1.3771
Epoch 1/2... Discriminator Loss: 1.2850... Generator Loss: 0.7463
Epoch 1/2... Discriminator Loss: 2.1356... Generator Loss: 0.2995
Epoch 1/2... Discriminator Loss: 1.2889... Generator Loss: 0.9628
Epoch 1/2... Discriminator Loss: 1.4516... Generator Loss: 1.0139
Epoch 1/2... Discriminator Loss: 1.6263... Generator Loss: 0.4897
Epoch 1/2... Discriminator Loss: 2.4224... Generator Loss: 2.6550
Epoch 1/2... Discriminator Loss: 1.4500... Generator Loss: 0.7272
Epoch 1/2... Discriminator Loss: 1.3284... Generator Loss: 0.8443

Epoch 1/2... Discriminator Loss: 1.4235... Generator Loss: 0.6116
Epoch 1/2... Discriminator Loss: 1.2477... Generator Loss: 1.2486
Epoch 1/2... Discriminator Loss: 1.4899... Generator Loss: 1.0452
Epoch 1/2... Discriminator Loss: 1.2501... Generator Loss: 0.9216
Epoch 1/2... Discriminator Loss: 3.6151... Generator Loss: 3.2790
Epoch 1/2... Discriminator Loss: 1.5153... Generator Loss: 0.8039
Epoch 1/2... Discriminator Loss: 1.4930... Generator Loss: 0.7968
Epoch 1/2... Discriminator Loss: 1.4394... Generator Loss: 0.7107
Epoch 1/2... Discriminator Loss: 1.3997... Generator Loss: 0.8662
Epoch 1/2... Discriminator Loss: 1.4496... Generator Loss: 0.5930

Epoch 1/2... Discriminator Loss: 1.2800... Generator Loss: 1.1714
Epoch 1/2... Discriminator Loss: 1.3432... Generator Loss: 0.9992
Epoch 1/2... Discriminator Loss: 1.2461... Generator Loss: 0.9732
Epoch 1/2... Discriminator Loss: 1.3892... Generator Loss: 0.7043
Epoch 1/2... Discriminator Loss: 1.2769... Generator Loss: 0.8955
Epoch 1/2... Discriminator Loss: 1.2158... Generator Loss: 0.7548
Epoch 1/2... Discriminator Loss: 1.4067... Generator Loss: 0.9350
Epoch 1/2... Discriminator Loss: 1.2916... Generator Loss: 0.8608
Epoch 1/2... Discriminator Loss: 1.4519... Generator Loss: 0.9774
Epoch 1/2... Discriminator Loss: 1.9347... Generator Loss: 0.2677

Epoch 1/2... Discriminator Loss: 1.3470... Generator Loss: 0.9458
Epoch 1/2... Discriminator Loss: 1.4457... Generator Loss: 0.7673
Epoch 1/2... Discriminator Loss: 1.4282... Generator Loss: 0.8004
Epoch 1/2... Discriminator Loss: 1.2651... Generator Loss: 0.9636
Epoch 1/2... Discriminator Loss: 1.4517... Generator Loss: 1.0940
Epoch 1/2... Discriminator Loss: 1.7676... Generator Loss: 1.8097
Epoch 1/2... Discriminator Loss: 1.4626... Generator Loss: 0.6496
Epoch 1/2... Discriminator Loss: 1.3309... Generator Loss: 0.6405
Epoch 1/2... Discriminator Loss: 1.4274... Generator Loss: 0.5556
Epoch 1/2... Discriminator Loss: 1.4895... Generator Loss: 0.7058

Epoch 1/2... Discriminator Loss: 1.3400... Generator Loss: 1.1656
Epoch 1/2... Discriminator Loss: 1.3887... Generator Loss: 0.6961
Epoch 1/2... Discriminator Loss: 1.2782... Generator Loss: 1.5053
Epoch 1/2... Discriminator Loss: 1.3870... Generator Loss: 0.8585
Epoch 1/2... Discriminator Loss: 1.4282... Generator Loss: 0.7382
Epoch 1/2... Discriminator Loss: 1.4689... Generator Loss: 0.8708
Epoch 1/2... Discriminator Loss: 1.1695... Generator Loss: 1.0266
Epoch 1/2... Discriminator Loss: 1.1962... Generator Loss: 1.2475
Epoch 1/2... Discriminator Loss: 1.2788... Generator Loss: 0.7688
Epoch 1/2... Discriminator Loss: 1.4464... Generator Loss: 0.4889

Epoch 1/2... Discriminator Loss: 1.3478... Generator Loss: 0.8385
Epoch 1/2... Discriminator Loss: 1.3692... Generator Loss: 0.6411
Epoch 1/2... Discriminator Loss: 1.4459... Generator Loss: 0.5092
Epoch 1/2... Discriminator Loss: 1.2021... Generator Loss: 0.9861
Epoch 1/2... Discriminator Loss: 1.2916... Generator Loss: 1.0627
Epoch 1/2... Discriminator Loss: 1.4255... Generator Loss: 0.7107
Epoch 1/2... Discriminator Loss: 1.6251... Generator Loss: 0.3502
Epoch 1/2... Discriminator Loss: 1.3805... Generator Loss: 0.5084
Epoch 1/2... Discriminator Loss: 1.5947... Generator Loss: 1.5417
Epoch 1/2... Discriminator Loss: 1.5021... Generator Loss: 0.8226

Epoch 1/2... Discriminator Loss: 1.4212... Generator Loss: 0.5085
Epoch 1/2... Discriminator Loss: 1.4596... Generator Loss: 0.5163
Epoch 1/2... Discriminator Loss: 1.0016... Generator Loss: 1.2501
Epoch 1/2... Discriminator Loss: 1.5512... Generator Loss: 0.9097
Epoch 1/2... Discriminator Loss: 1.5068... Generator Loss: 0.9199
Epoch 1/2... Discriminator Loss: 1.5302... Generator Loss: 0.7051
Epoch 1/2... Discriminator Loss: 1.3830... Generator Loss: 0.6475
Epoch 1/2... Discriminator Loss: 1.4186... Generator Loss: 0.7031
Epoch 1/2... Discriminator Loss: 1.4293... Generator Loss: 0.8144
Epoch 1/2... Discriminator Loss: 1.3962... Generator Loss: 0.9153

Epoch 1/2... Discriminator Loss: 1.4341... Generator Loss: 0.6941
Epoch 1/2... Discriminator Loss: 1.4198... Generator Loss: 0.8501
Epoch 1/2... Discriminator Loss: 1.4548... Generator Loss: 0.8949
Epoch 1/2... Discriminator Loss: 1.3904... Generator Loss: 0.6376
Epoch 1/2... Discriminator Loss: 1.5006... Generator Loss: 0.6742
Epoch 1/2... Discriminator Loss: 1.3502... Generator Loss: 0.7797
Epoch 1/2... Discriminator Loss: 1.4867... Generator Loss: 0.5660
Epoch 1/2... Discriminator Loss: 1.4211... Generator Loss: 0.8188
Epoch 1/2... Discriminator Loss: 1.3994... Generator Loss: 0.7476
Epoch 1/2... Discriminator Loss: 1.4162... Generator Loss: 0.7489

Epoch 1/2... Discriminator Loss: 1.4075... Generator Loss: 0.6389
Epoch 1/2... Discriminator Loss: 1.3805... Generator Loss: 0.6967
Epoch 1/2... Discriminator Loss: 1.4734... Generator Loss: 0.8085
Epoch 1/2... Discriminator Loss: 1.4297... Generator Loss: 0.7913
Epoch 1/2... Discriminator Loss: 1.3931... Generator Loss: 0.7066
Epoch 1/2... Discriminator Loss: 1.3818... Generator Loss: 0.7484
Epoch 1/2... Discriminator Loss: 1.3920... Generator Loss: 1.0200
Epoch 1/2... Discriminator Loss: 1.3832... Generator Loss: 0.7032
Epoch 1/2... Discriminator Loss: 1.3844... Generator Loss: 0.8797
Epoch 1/2... Discriminator Loss: 1.3530... Generator Loss: 0.8946

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.